Digitalisation of shipyard production planning: a review of simulation, optimisation, AI, and digital twin methods (2010–2025)

Digitalisation is reshaping shipyard production, yet its methodological foundations remain fragmented across simulation, optimisation, Artificial Intelligence (AI), and Digital Twin (DT) research streams. This paper presents a domain-specific methodological review of shipyard production modelling from 2010 to 2025, synthesising advances in Discrete-Event Simulation (DES), multi-objective optimisation, hybrid simulation–optimisation architectures, Machine Learning (ML), reinforcement learning (RL), and DT-enabled cyber-physical systems. Using an explicit evaluative framework based on integration depth, validation basis, and decision scope, the review differentiates between analytically mature but execution-decoupled DES/optimisation approaches and integration-rich yet variably validated DT and AI-driven systems. The analysis shows that hybrid DES-optimisation frameworks currently represent the most operationally credible class of methods, delivering measurable production improvements under structured conditions, whereas many DT and AI contributions prioritise architectural integration and data synchronisation over longitudinal yard-wide KPI validation. A comparative assessment of simulation platforms, optimisation engines, and manufacturing execution system/enterprise resource planning/product lifecycle management infrastructures highlights the central role of structured product–process–resource data and execution-layer connectivity, while severe confidentiality constraints and the scarcity of openly available industrial datasets continue to limit reproducibility and benchmarking. Overall, shipyard production research is progressing toward increasingly integrated and cyber-physical systems, but sustained yard-scale validation and shared benchmark development remain critical prerequisites for translating architectural sophistication into demonstrable operational impact.